Optimization and adaptation of situation-aware applications based on workflow fragments

点击翻译按钮获取中文摘要

The solution SitOPT, which has been developed in the course of the project so far, enables dynamic, autonomous adaptation of workflows to present situations. In a dynamic environment, e.g. a production environment, it is possible to adapt a workflow execution in an appropriate manner to changing and possibly unforeseen situations. SitOPT is pursuing a centralized approach to the control of dynamic environments, which is not suitable for all application domains.In the continuation application, the concept DiStOPT (Distributed SitOPT) is to be developed, in which the developed concepts are extended. The objective of DiStOPT is to monitor distributed dynamic environments on a decentralized basis, in order to adapt processes, which involve several participants and are modeled in the form of choreographies, automatically and globally at runtime. For example, Internet of Things or Industry 4.0 environments can be efficiently monitored and related workflows can be adapted autonomously. Scalability, availability and near-real-time processing play a fundamental role in these distributed scenarios and must therefore be covered by new concepts that clearly go beyond the mechanisms of the previous project.The objective of the expanded research project DiStOPT is to develop concepts and methods that allow decentralized monitoring and adaptation of distributed dynamic environments. In particular, processes that involve more than one participant should be automatically adapted in a global perspective at runtime depending on the situation. In order to monitor modern Internet of Things or Industry 4.0 environments with high efficiency requirements, a flexible deployment of the SitOPT components is designed to detect occurring situations in distributed, diverse system environments. These include, for example, various cloud types, such as edge cloud, private cloud, or public cloud. An edge cloud refers to a system environment that has very short communication paths to context-aware sensors. As a result, data can be processed as close as possible to the sensor and thus long communication paths can be avoided. This allows near-real-time processing of the data to provide situations. Non-time-critical, computationally intensive calculations can, however, be outsourced in private or public-cloud environments. In order to achieve this, a multi-cloud-based management of the SitOPT components on distributed systems must be created. In order to adapt the processes in a collaborative manner, concepts for situation-adaptive choreographies have to be developed which bring the workflow adaptation to a higher level. Both local situations that occur with individual partners as well as global situations affecting all participating partners must be considered.